Don't lose your points!
Sign up and save them.
Matched-pair Design

Matched-pair Design

Author: Jonathan Osters

This lesson will explain matched-pair design experiments

See More

Try Our College Algebra Course. For FREE.

Sophia’s self-paced online courses are a great way to save time and money as you earn credits eligible for transfer to many different colleges and universities.*

Begin Free Trial
No credit card required

28 Sophia partners guarantee credit transfer.

281 Institutions have accepted or given pre-approval for credit transfer.

* The American Council on Education's College Credit Recommendation Service (ACE Credit®) has evaluated and recommended college credit for 25 of Sophia’s online courses. Many different colleges and universities consider ACE CREDIT recommendations in determining the applicability to their course and degree programs.


Matched-Pair Design

Video Transcription

Download PDF

This tutorial is going to talk to you about matched-pair design. This is also sometimes called matched-pairs.

So suppose that you have a tire company that's considering rolling out a new type of rubber for its bicycle tires-- pun completely intended. It has 300 bicycles available. And in a completely randomized design, on which there's another tutorial, the researchers would place the numbers 1 through 300 in a hat and have the bikers who pull the numbers 1 through 150 receive the old rubber tires, and the 151 through 300-- anyone who pulls those-- receive the new rubber tires. And the bikers don't necessarily need to know who's getting which tires.

However, the problem with this is what if the 300 riders don't all ride the same way or equally as often? What do you do then? How do we create two groups that are roughly the same, with the exception of the bicycle tires?

Well, one way to do it is with a matched-pairs design, and it looks like this. A preferable design would be matched-pairs. So the researchers could still put the numbers 1 through 300 in a hat. The only difference is the people who pull out 1 through 150 would get both the old and the new. They would put the old in the front and the new rubber tire in the back. And the people who pulled out 151 through 300 would get the new rubber tire in the front and the old one in the back.

So there's still some randomization going on. The only difference is every biker is going to get one old tire and one new tire. And this is going to allow us to compare the tread wear for each bike, because the front tire and the rear tire both get worn at approximately equally the same way. And it doesn't matter if the biker rides a lot or rides a little. We can directly compare the old to the new or the old to the new.

And so a matched-pairs design is matching subjects into groups of two that are as similar as possible with respect to any variable we think might affect the outcome. In this case, the amount that the bikers rode their bikes might affect the tread wear, which is what we were interested in. And so what we did was we matched up each individual's bike tires. And so each rider got to ride on both types of tire.

And the vast majority of the time, this is exactly what we do with a matched-pairs design. We assign each subject to both groups. Instead of assigning them to one group, we assign everyone to both groups and randomly assign the order in which those treatments are applied. And each participant then counts as his or her own matched pair.

Now, that's not always how it's done. That's typically how a matched-pairs design is done, where you essentially compare someone to themselves. However, we could do a matched pairs design in a case like this.

Suppose you had 20 participants for an experiment for a flu vaccine. And we think that maybe gender and age will play a role in how well this treatment works. So what we think we might want to do is create groups of two that are as similar as possible with respect to any variable that we think might affect the outcome. So here we have 10 men and 10 women-- all of these different ages. So how can we create groups that are as similar as possible?

One way to do it would be to first list the participants by gender. So participant 1, 3, 4, 8, 10, 11, 12, 16, 18, and 20 are the males. And these are their ages. And these are the females and their ages. Well, what can we do now?

Well, we think that age also plays a role. So what we can do is, within the male category, pick the two ages that we think are the closest together-- 24 and 25. So participants 1 and 8 will form a matched pair. We can continue going along with that logic by saying participants 1 to 8-- the 24 and 25-year-old males-- will form a matched pair. Participants 10 and 12, 4 and 3, 20 and 11, and 16 and 18-- these are similarly aged males, and we can apply the same criteria for similarly aged females. And so these groups of two are your matched pairs-- 1 and 8, 2 and 7, 12 and 10, 14 and 17, and so on. Now, to continue the experiment, what we should do is randomly assign one of these two participants-- 1 or 8-- to the flu vaccine and the other one to the control group.

So to recap, in a matched-pairs design, two numbers whose characteristics are very similar are paired, then each one is sent to a different group. This can also happen with one member, and obviously one member's characteristics are very similar to himself. Matched-pairs designs are often done by assigning both treatments to every participant. And this is a very common use of the matched-pairs design.

Good luck, and we'll see you next time.

Terms to Know
Matched-Pair Design

An experimental design where two subjects who are similar with respect to variables that could affect the outcome of the experiment are paired together, then one of them is assigned to one treatment and one is assigned to the other. This can also be done by assigning each subject to both treatments, where each subjects acts as their own matched-pair.